About

Short Bio

(Sòng Jiàn)

Hello, I am Jian Song, a first-year Computer Science PhD student advised by Professor Zhengxiong Li at the University of Colorado Denver.

I am originally from Xinzheng, Henan — a city rich in history as the legendary birthplace of the Yellow Emperor and the former capital of the Zheng and Han states. It now serves as a major modern transit hub, housing the Zhengzhou Xinzheng International Airport.

Research Interests

University of Colorado Denver (2025 – present, advised by Prof. Zhengxiong Li)

  • Physical-space AI & Robotics: Developing robotics-oriented AI systems that operate in and reason about the physical world, with a focus on tactile perception and multi-modal sensory integration for embodied intelligence.

Westlake University (2023 – 2025, with Prof. Donglin Wang)

  • Neuromorphic Computation: Utilizing ANN-SNN hybrid training and ANN2SNN conversion to optimize the balance between performance and energy consumption of RL agents across multiple modalities (images, text, 3D data).
  • Event-based Vision: Improving transformer-based recognition on DVS and RGB datasets for low-latency, low-power autonomous driving.
  • VLM in Robotic Learning: Integrating vision, 3D geometric data, and language for comprehensive robotic environment understanding.

Publications

  1. Activation-wise Propagation: A One-Timestep Strategy for Spiking Neural Networks AAAI'26
    Activation-wise Propagation: A One-Timestep Strategy for Spiking Neural Networks
    Jian Song, Xiangfei Yang, Shangke Lyu, Donglin Wang
    AAAI Conference on Artificial Intelligence (AAAI).
    Singapore, 2026
    Paper Code

  2. Adaptive Spiking TD3+BC for Offline-To-Online Spiking Reinforcement Learning IJCNN'24
    Adaptive Spiking TD3+BC for Offline-To-Online Spiking Reinforcement Learning
    Xiangfei Yang, Jian Song, Xuetao Zhang, Donglin Wang
    International Joint Conference on Neural Networks (IJCNN).
    Yokohama, Japan, 2024
    Paper

  3. Calibrating the Converted Spiking Reinforcement Learning ICIC'24
    Calibrating the Converted Spiking Reinforcement Learning
    Jian Song, Xiangfei Yang, Xuetao Zhang, Donglin Wang
    International Conference on Intelligent Computing (ICIC).
    Tianjin, China, 2024
    Paper Oral

Projects

  • Gait Customization for WAVEGO: Developed various gaits and interactive actions for WAVEGO, a 12-DOF micro quadruped robot powered by ESP32 sub-controller and Raspberry Pi 4B host controller.
  • Trajectory Prediction of Fruit Fly’s Neural Visual Motor System: Conducted research on predicting the trajectory of a fruit fly’s neural visual motor system, aiming to understand the underlying mechanisms of visual-motor integration in neural systems.
  • Review of Neuromorphic Imitation Learning: Conducted a comprehensive review of neuromorphic imitation learning, implementing spiking neural networks (SNN) in imitation learning baselines such as GAIL, AIRL, SQIL, and IQ Learn.
  • 3D Reconstruction of Nucleosome: Researched the flexible regions of nucleosome combination while developing methods to prevent noise interference in the reconstruction process.

Educations

  • 2025.08 - present, Ph.D. in Computer Science, University of Colorado Denver, Colorado, USA.
  • 2021.09 - 2022.09, M.Sc in Drug Discovery and Translational Biology with Merit, The University of Edinburgh, Scotland, UK.
  • 2017.09 - 2021.06, B.Sc in Biotechnology, Central China Normal University, Wuhan, China.

Email

  • jian.song [AT] ucdenver [DOT] edu